Pathological Changes of Small Vessel Disease in Intracerebral Hemorrhage: a Systematic Review and Meta-analysis

In intracerebral hemorrhage (ICH) with pathology-proven etiology, we performed a systematic review and meta-analysis to elucidate the association between cerebral amyloid angiopathy (CAA) and arteriolosclerosis, and directly compared MRI and pathological changes of markers of cerebral small vessel disease (CSVD). Studies enrolling primary ICH who had received an etiological diagnosis through biopsy or autopsy were searched using Ovid MEDLINE, PubMed, and Web of Science from inception to June 8, 2022. We extracted pathological changes of CSVD for each patient whenever available. Patients were grouped into CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis subgroups. Of 4155 studies identified, 28 studies with 456 ICH patients were included. The frequency of lobar ICH (p<0.001) and total microbleed number (p=0.015) differed among patients with CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis. Concerning pathology, severe CAA was associated with arteriolosclerosis (OR 6.067, 95% CI 1.107–33.238, p=0.038), although this association was not statistically significant after adjusting for age and sex. Additionally, the total microbleed number (median 15 vs. 0, p=0.006) was higher in ICH patients with CAA evidence than those without CAA. The pathology of CSVD imaging markers was mostly investigated in CAA-ICH. There was inconsistency concerning CAA severity surrounding microbleeds. Small diffusion-weighted imaging lesions could be matched to acute microinfarct histopathologically. Studies that directly correlated MRI and pathology of lacunes, enlarged perivascular spaces, and atrophy were scarce. Arteriolosclerosis might be associated with severe CAA. The pathological changes of CSVD markers by ICH etiology are needed to be investigated further. Supplementary Information The online version contains supplementary material available at 10.1007/s12975-023-01154-4.


Introduction
Primary intracerebral hemorrhage (ICH) results in high morbidity and disability.Cerebral small vessel disease (CSVD) is the major cause of primary ICH [1]: arteriolosclerosis, which is closely associated with hypertension [2], mainly accounts for deep ICH, and cerebral amyloid angiopathy (CAA) accounts for lobar ICH.Notably, CAA and arteriolosclerosis coexist frequently.Previous studies have found that lobar ICH/microbleed and deep ICH/microbleed could coexist in a patient [3].Additionally, animal studies have found that an increase in systolic blood pressure promotes the occurrence of ICH in Tg2576 mice characterized by amyloid beta (Aβ) deposition [4].And in a clinical study, subgroup analysis of the PROGRESS study has shown that antihypertensive treatment (vs.placebo) reduced the risk of ICH in CAA patients [5].However, there is currently a lack of systematic reviews that assess the association between Mangmang Xu and Yuyi Zhu contributed equally.
CAA and arteriolosclerosis in ICH patients with pathologyproven evidence.
On the other hand, increasing evidence has demonstrated that imaging markers of CSVD are related to the presence and prognosis of primary ICH [6,7].However, the characteristics of CSVD markers in patients with pathological evidence for arteriolosclerosis-and CAA-related ICH, and the underlying pathological changes of theses markers, remain unclear.Although a previous study [8] reviewed the pathological substrates of CSVD such as cerebral microbleeds (CMB) and white matter changes, the majority of included studies investigated Alzheimer's disease.Despite 80~97% of patients with Alzheimer's disease have CAA at autopsy, most of them do not suffer from symptomatic ICH [9,10].We, therefore, performed a systematic review and meta-analysis to elucidate (1) the association between arteriolosclerosis and CAA and their imaging characteristics, and (2) the imaging and pathological correlations of CSVD markers in primary ICH patients to better understand the mechanisms underlying primary ICH.

Search Strategy
Our study was approved by the Ethics Committee on Biomedical Research, West China Hospital of Sichuan University, and was registered in PROSPERO (CRD42022343347).We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses [11] guidelines.We systematically searched PubMed, Ovid Medline, and Web of Science for studies investigating the neuropathology of CSVD in patients with primary ICH until 8 June 2022.The following terms were used: "(cerebral hemorrhage) OR (intracranial hemorrhage)" AND "(patholog*) OR (postmortem) OR (autopsy) OR (biopsy)" AND "(cerebral small vessel diseases) OR (microvessel) OR (white matter) OR (microinfarct) OR (lacune) OR (perivascular space) OR (microbleed) OR (microhemorrhage) OR (atrophy)."Additionally, references to relevant studies were searched.The details of the study selection process are presented in Fig. 1.

Eligibility Criteria
The included criteria were as follows: investigating adult patients (age≥18 years) with primary ICH; having autopsy or biopsy to confirm the etiology for arteriolosclerosis and CAA; with or without CSVD markers, including white matter changes, CMB, cerebral microinfarcts (CMI), lacunes, enlarged perivascular spaces (EPVS), brain atrophy, and the total CSVD burden.Abstract, review, editorial, and basic research were excluded.This systematic review only focused on primary ICH, not on hemorrhage into the subarachnoid space, epidural space, or subdural space.Pure intraventricular hemorrhage, ICH secondary to trauma, structural vascular lesions, inflammation, infections, inherited or genetic small vessel diseases, and hemorrhagic transformation after ischemic stroke were beyond the scope of this review.We only included studies that were published in English.

Data Extraction
Two authors (Mangmang Xu and Xindi Song) independently searched the literature and extracted data.Any disagreement was resolved by consensus.A data extraction form was developed using SPSS.For each included study, we extracted the family name of the first author, publication year, country of origin, the sample size of primary ICH, examination method (autopsy or biopsy), patient demographics (male sex, age at death or biopsy), pathology diagnosis (CAA or arteriolosclerosis), ICH location (lobe, deep, cerebellum), and CSVD imaging markers including white matter changes, lacunes, CMB, CMI, EPVS, atrophy, and total CSVD burden.For studies that investigated ICH with CAA evidence, we additionally extracted information on the diagnostic method of CAA, CAA-related small vessel wall changes (SVWCs), the coexistence of arteriolosclerosis, senile plaques, and blood vessel changes in or around the hematoma.

Definitions
Lobar ICH included hemorrhages originating from the frontal, temporal, parietal, or occipital lobes of the brain.For the two studies [12,13] that enrolled clinically diagnosed CAA-ICH according to the modified Boston criteria without reporting ICH location, we regarded those cases as having lobe ICH.Deep ICH was considered when originating from the basal ganglia, thalamus, internal capsule, external capsule, or brain stem.Biology included brain biopsy and biopsy at hematoma evacuation in this study.
Severe CAA in histopathology was defined as follows: (1) a CAA score of 9-12 for the study [12] that used a cumulative cortical CAA score with a range of 0-12 by calculating the basis of Aβ stained sections from four areas, (2) a CAA score of 3-4 for those studies [14][15][16][17] that used the grading scale by Greenberg and Vonsattel with a range of 0-4, (3) a CAA score of 3+ indicating that many lesions were visible [18] in a scale of 1+~3+, or as per the original study authors' definition [19][20][21][22].

Statistical Analysis
Most of the included studies were case reports, so we extracted individual patient data from those cases.The demographic data in the study by Ter Telgte [12] were the same as that in the study by van Veluw [13]; therefore, we only extracted data from the latter.For CMB analysis, we extracted CMB number in ex vivo MRI in the study by van Veluw 2016 [17], under the other two studies [13,20].Independent-samples T test, one-way ANOVA, Mann-Whitney U test, or Kruskal-Wallis H test were used for continuous variables, and Pearson chi-squared test or Fisher's exact test for categorical variables, when appropriate.Multivariate analysis was performed using binary logistic regression.For the statistical analysis in our present study, hemorrhage in the cerebellum was categorized as lobar ICH.All statistical analyses were performed using IBM SPSS Statistics (version 23), and a p value of ≤0.05 was considered statistically significant.

Study Characteristics
A total of 4155 citations and 509 full texts were screened (Fig. 1).Of these, 28 studies with 456 participants were included .The characteristics of included studies are shown in Table 1.Most of the studies were conducted in the USA (39.3%, 11/28), followed by Japan (28.6%, 8/28).Three hundred twenty patients had CAA with or without arteriolosclerosis; 48 patients had strict CAA while 19 had CAA + arteriolosclerosis.CAA-related SVWCs were investigated in 17 studies [14-22, 24, 29-33, 38, 39], and their detailed pathological changes are summarized in the Supplementary table.
For patients with ICH due to strict arteriolosclerosis, the majority (61.1%) of ICH were located in deep regions of the brain.The presence of arteriolosclerosis was significantly associated with hypertension (percentage of hypertension in arteriosclerosis vs. non-arteriolosclerosis: 74.6 vs. 36.4%,p=0.001).This association remained significant after correcting for age and sex (adjusted OR 3.329, 95% CI 1.056-10.495,p=0.040).

White Matter Changes
Overall, six studies [15,18,20,30,32,36] investigated white matter changes on CT or MRI.Of these, three studies [18,20,30] examined the correlation between MRI and pathological changes in white matter lesions (Table 3).White matter changes were associated with rarefaction of myelin staining, diffuse or patchy myelin loss, diffuse white matter edema, and even advanced white matter loss [20,30].Histologically, the white matter appeared vacuolated and was accompanied by swollen oligodendrocytes and astrocytic proliferation [18,30].Amyloid plaques that were positive for Aβ were occasionally observed in the white matter [18].However, CAA-associated vasculopathy did not always result in ischemic white matter lesions [18].

Cerebral Microinfarct
Four studies [12,13,17,25] investigated CMI, and all four investigated pathological changes of CMI, as detailed in Table 4.The findings suggested that small DWI lesions could be matched to acute CMI histopathologically, characterized by tissue pallor consisting of eosinophilic neurons, absent or only mild reactive astrocytes, and absent or fragmented microglia.And chronic cortical CMI was associated with tissue loss or central cavitation, as well as many reactive astrocytes around the lesions and few reactive or amoeboid microglia [12].CAA was found to be severe surrounding a CMI, particularly for CMI in those with lobar ICH.

Lacunes, EPVS, and Brain Atrophy
Each of the three imaging markers was investigated in a separate study (Table 1).Patients with brain atrophy corresponded to amyloid deposition in the parenchyma and small vessels [15].However, the pathological changes of lacune and EPVS in primary ICH were not investigated.Twenty-one of 34 CMBs on MRI were confirmed pathologically, characterized by focal accumulation of hemosiderin-laden macrophages, and sometimes tissue necrosis.Hemosiderin deposits (often smaller and contained lesser hemosiderin-laden macrophages) were observed in the brains which were negative for CMB on MRI.Hernandez-Guillamon 2012 [22] Fifteen chronic bleeds containing hemosiderin or other iron storage complexes were found in the lobes, among which 7 were positive for Aβ staining.van Veluw 2019 [13] The rupture site of vessel contained less Aβ, less intact smooth muscle cells, and extensive fibrin.The upstream or downstream of the rupture site contained more Aβ.The involved vessels of CMB were enlarged and fibrinoid necrotic, with extensive vessel wall remodeling.Guidoux 2018 [35] Forty-five of 48 CMBs on MRI were confirmed pathologically, corresponding to small bleeds with lytic red bloods cells for recent CMB, and macrophages for old CMB.There was no significant difference in CMB size on pathology between those with and without CAA.Greenberg 2009 [34] Patients with high CMB count (>50) had increased thickness of amyloid-positive vessel wall than those with few CMBs (<3).Shelton 2015 [31] No CMB on neuroimaging.Autopsy showed hemosiderin deposition with spongiosis in perivascular regions.Jolink 2022 [25] Total CMB number was higher in lobar ICH than non-lobar ICH, without reaching a significant difference.Cortical CMB were more frequently noted in superficial layers of cortex in lobar ICH, and in deep layers of cortex in non-lobar ICH.CAA was severer surround the CMB on lobar ICH than that in non-lobar ICH.Severe CAA of the involved vessels of CMB was infrequently observed.van Veluw 2016 [17] Ex vivo MRI detected 171 CMBs (all located in cortical ribbon), compared to ~66 in vivo MRI.Acute CMBs were characterized by accumulation of intact red blood cells.Old or subacute CMBs corresponded focal hemosiderin deposits.Schrag 2010 [16] Seven, 2, and 2 CMBs in cortical grey matter, deep grey matter, and white matter, respectively.Susceptibility-weighted imaging overestimated CMB size.CMI Ter Telgte 2020 [12] Acute cortical CMI: tissue pallor consisting of eosinophilic neurons, absent or mild reactive astrocytes, and absent or fragmented microglia.Chronic cortical CMI: tissue loss or central cavitation, with many reactive astrocytes and few reactive or amoeboid microglia.van Veluw 2019 [13] Many Aβ-positive vessels surrounding a CMI.Mild fibrin and absent or only few intact smooth muscle cells in the vessels involved in CMI; the involved vessels were intact but stiff due to the deposition of Aβ, with relatively narrow lumens.Jolink 2022 [25] More CMI in lobar ICH than non-lobar ICH.Cortical CMI were more frequently noted in superficial layers of the cortex in lobar ICH, and in deep layers of the cortex in non-lobar ICH.CAA was severer surround the CMI in lobar ICH than that in non-lobar ICH.Severe CAA of the involved vessels of CMI was infrequently observed.van Veluw 2016 [17] CMI tended to cluster, characterized by ischemic or shrunken neurons for acute CMI and tissue loss and gliosis for chronic CMI.

Lacunes
Fazekas 1999 [20] Five of 7 patients with MRI lacunes in the basal ganglia and thalami were histologically confirmed.

CSVD burden
Charidimou 2016 [36] Higher CSVD score was associated with CAA-related abnormities on pathology and the presentation of ICH.Pasi 2020 [26] Five of 8 patients could be rated for CSVD score.Deep ICH were scored 1 for moderate arteriosclerosis, and 3 or 4 for mild arteriosclerosis.Lobar ICH without arteriosclerosis were scored 2.

Total CSVD Burden
Two studies [26,36] investigated the total CSVD burden (Table 4).A direct histopathological-MRI study demonstrated that the total CAA burden of CSVD, as determined by summing the four most characteristic MRI markers of CAA, was associated with the presentation of symptomatic CAA-related ICH (versus CAA without ICH) and CAArelated microangiopathy on pathology [36].

Discussion
We presented here a comprehensive systematic review of published articles that investigated ICH with pathologyproven evidence for arteriolosclerosis and CAA.Our study directly compared MRI and pathology to clarify the pathological changes of CSVD imaging markers.To the best of our knowledge, this is the first review to assess the interaction between arteriolosclerosis and CAA by summarizing published cases with a pathology diagnosis of CAA or arteriolosclerosis in ICH.Our findings provide pathological evidence that there might be differences in lobar ICH and total microbleed number among CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis, and there might be an association between arteriolosclerosis and severe CAA, which needs to be investigated further in future studies.
Although the SVWCs in CAA-ICH were found to be similar to those in CAA without ICH, in terms of the location of Aβ deposition and the changes in the lumen [40], our findings suggest that severe CAA (usually characterized by double barreling and/or fibrinoid necrosis) is highly prevalent in CAA-ICH.Our results, together with the findings of Vonsattel et al. [21], indicate that a severe degree of CAA represents an important feature of CAA-ICH [21].
Regarding the CSVD markers in ICH, we found that patients with CAA had higher number of total CMB, particularly for patients with strict CAA, as compared to those without CAA on pathology.This finding was in line with previous study by Edip Gurol et al. that found that patients with CAA-ICH had higher number of total CMB as compared to those with non-CAA related ICH [41].In addition, the total CAA burden assessed in neuroimaging was associated with the presentation of symptomatic CAA-related ICH and CAA-related microangiopathy on pathology.Therefore, it might be reasonable to speculate that higher individual CSVD markers indicate greater Aβ severity.However, our analysis of individual patient data showed that severe CAA was not associated with CMB number, which is in line with the findings of another study that found CMB to be frequently located around Aβ negative small vasculature [42].In addition, one amyloid positron emission tomography study evaluating CAA burden in both the ICH-affected hemisphere and the ICH-free hemisphere indicated that amyloid burden was similarly distributed and ICH was unlikely to be directly linked to amyloid burden [43].Taken together, these findings challenge the hypothesis that the ruptured vessel reflects a significant Aβ load [42].
As expected, hypertension was significantly associated with arteriolosclerosis, which was highly prevalent in CAA-ICH as well.It is widely accepted that lobar hemorrhage is required for CAA diagnosis per the Boston Criteria [44].In this study, we found that all cases of strict CAA-ICH were located in the lobes.However, we also found that in some patients with the coexistence of CAA and arteriolosclerosis on pathology, the hemorrhage could be located in deep territories, suggesting that the presence of deep ICH could not preclude the possibility of amyloid angiopathy in the elderly patients.
Overall, the pathology of CSVD markers in ICH was under-researched, with most studies focusing on CAA-ICH.While the majority of CMBs observed on MRI could be confirmed histopathologically, several CMBs could not be confirmed, with no abnormality or small cavities at neuropathological amination [20,35].Compared with standard histopathological sections, MRI tends to underestimate the total burden of CMB [20], as well as CMI [13].The lesions that could not be retrieved for histological analysis were generally too small [25].Therefore, future studies with highfield MRI scanners are warranted to capture the total burden of these two markers [20,35].Of note, CMB on MRI was traditionally recognized to represent foci of past hemorrhage [20,22].However, this review found that CMB could also correspond to recent microhemorrhage with small bleeds in patients with ICH [35].
Another interesting finding was the inconsistency in CAA severity around CMB [13,25].This inconsistency might be attributed to the different methodologies used by the two studies.Jolink et al. [25] investigated CAA severity for each patient by taking samples with multiple lesions, while van Veluw et al. [13] assessed CAA for each section that contained CMB.Studies that directly correlated MRI and pathology of lacunes, EPVS, and atrophy were scarce.Thus, future studies are warranted to further elucidate the CAA pathology around CMB and CMI, and to assess the pathology of lacunes, EPVS, and atrophy in ICH.

Strength and Limitation
The present systematic review provided the most recent and comprehensive overview of the pathology of CSVD in primary ICH.Studies were comprehensively searched from databases using a systematic search strategy, as well as from handsearching of references of relevant studies.However, the majority of included studies only assessed the presence of CAA on pathology, without information on arteriolosclerosis, resulting in a relatively small number of cases with CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis.Despite this, our study has the largest number of ICH cases with pathology-proven etiology across the literature and firstly provides the clinical and imaging information for CAA+arteriolosclerosis, strict CAA, and strict arteriolosclerosis.

Conclusion and Considerations for Future Research
In summary, our study provides synthesized pathological evidence for the hypothesis that there might be an interaction between CAA and arteriolosclerosis.Future studies are needed to verify our findings.The pathological changes of CSVD features in neuroimaging are heterogeneous, and MRI-histopathological correlation studies of lacunes and EPVS remain scarce.Therefore, further investigation is needed to explore the pathological changes of CSVD markers by ICH etiology.Additionally, studies using advanced MRI techniques with a high spatial resolution to visualize structural and functional brain changes of CSVD in ICH are also warranted.

Fig. 1
Fig. 1 Flow chart of study selection

Table 1
Characteristics of included studies which investigating primary ICH ‡ Nine of 12 patients in this study had ICH.The age and male proportion were derived from the 12 patients § Sixteen of 20 patients in this study had primary ICH with pathology-based etiology || The original study included 20 patients at surgery and 16 brains at autopsy, but 21 in total went to the investigation of small vessel walls #The original study included 17 CAA-ICH, among which 12 were sporadic **Six of 46 patients had primary ICH with pathology evidence † † Six of 7 patients in this study had primary ICH.‡ ‡ Five of 6 patients in this study had primary ICH § § Ten of 12 patients in this study had ICH |||| One of 8 patients in this study had primary ICH ##Six of 12 patients in this study had primary ICH.CAA-ICHs in the two studies overlapped ***One of 5 patients in this study had ICH

Table 2
The clinical characteristics by CAA severity and ICH etiology *Eighteen and 49 patients in non-severe and severe group had data on hypertension, respectively † Three, 5, and 2 patients in CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis group had data on CMI number, respectively ‡ Sixteen and 12 patients in non-severe and severe group had data on arteriolosclerosis, respectively § Twelve, 30, and 16 patients in CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis group had data on hypertension, respectively || Four and 3 patients in non-severe and severe group had data on CMI number, respectively # Both severe and non-severe groups had 5 patients with data on total CMB number **Five, 5, and 11 patients in CAA + arteriolosclerosis, strict CAA, and strict arteriolosclerosis group had data on CMB number, respectively

Table 3
Neuropathological changes of white matter changes on imaging